Antibodies are proteins produced by lymphocytes known as B cells in vertebrates in response to stimulation by antigens. The basic structural unit of an antibody (a.k.a. immunoglobulin (Ig)) molecule consists of four polypeptide chains which come together in the shape of a capital letter “Y”. Two of the four chains are identical light (L) chains and two are identical heavy (H) chains. There are five different kinds (isotypes) of heavy chains which divide antibodies into five classes, namely, IgA, IgD, IgE, IgG and IgM. In addition, there are two different isotypes of light chains designated K and λ. Each class of heavy chains can combine with either of the light chains. The heavy and light chains each contain a variable region (VH and VL, respectively) that is involved in antigen binding and a constant (C) region. The antigen binding site is composed of six hypervariable regions (a.k.a. complementarity determining regions (CDRs)). Three CDRs from the heavy chain and three CDRs from the light chain are respectively positioned between four relatively conserved anti-parallel ∃-sheets which are called framework regions (FR1, FR2, FR3 and FR4), on each chain. By convention, numbering systems have been utilized to designate the location of the component parts of VH and VL chains. The Kabat definition is based on sequence variability and the Chothia definition is based on the location of structural loop regions.
For each type of Ig chain synthesized by B cells, there is a separate pool of gene segments, known as germline genes, from which a single polypeptide chain is synthesized. Each pool is located on a different chromosome and typically contains a relatively large number of gene segments encoding the V region and a lesser number of gene segments encoding the C region. Each light chain V region is encoded by a nucleic acid sequence assembled from two kinds of germline gene segments, i.e., a long V gene segment, a short joining (J) gene segment, and a C segment. The heavy chain is encoded by four kinds of germline gene segments, three for the variable region and one for the constant region. The three germline gene segments that encode the heavy chain variable region are a V segment, a J segment and a diversity (D) segment. Human germline V, D and J gene sequences have been characterized. The human germline VH gene segments (such “segments” are also referred to herein as family members) are classified into seven families (VH1-VH7) based on sequence homology of at least 80%. See, e.g., Matsuda, et al. J. Exp. Med. (1998) 188:2151-2162. There are approximately fifty-one VH segments (family members). The first two CDRs and three framework regions of the heavy chain variable region are encoded by VH. CDR3 is encoded by a few nucleotides of VH, all of DH and part of JH, while FR4 is encoded by the remainder of the JH gene segment. With regard to light chains, V Kappa (Vk) or V lambda (Vλ) gene segments (family members) encode the first two CDR and three framework regions of the V region along with a few residues of CDR3. J Kappa (JK) and J Lambda (Jλ) segments encode the remainder of the CDR3 region in a Vk or Vλ region, respectively. DNA encoding the k chain includes approximately forty Vk segments (family members) that are classified into six families (Vk I-Vk VI) based on sequence homology. DNA encoding the λ chain includes approximately thirty-one Vλ segments (family members) that are classified into ten families. See FIGS. 1, 2, 3 and 6.
Antibodies and antibody fragments have become promising therapeutic agents in connection with various human diseases in both acute and chronic settings. There are several methods being utilized to generate antibodies including hybridoma technology, bacterial display, ribosome display, yeast display, and recombinant expression of human antibody fragments on the surface of replicative bacteriophage. Monoclonal antibodies (mAbs), which may be produced by hybridomas, have been applied successfully as diagnostics for many years, but their use as therapeutic agents is just emerging. The vast majority of mAbs are of non-human (largely rodent) origin, posing the problem of immunogenicity in humans. When antibodies of rodent origin are administered to humans, anti-rodent antibodies are generated which result in enhanced clearance of the rodent antibody from the serum, blocking of its therapeutic effect and hypersensitivity reactions. These limitations have prompted the development of engineering technologies known as “humanization”.
The first humanization strategies were based on the knowledge that heavy and light chain variable domains are responsible for binding to antigen, and the constant domains for effector function. Chimeric antibodies were created, for example, by transplanting the variable domains of a rodent mAb to the constant domains of human antibodies (e.g. Neuberger M S, et al., Nature 314, 268-70, 1985 and Takeda, et al., Nature 314, 452-4, 1985). Although these chimeric antibodies induce better effector functions in humans and exhibit reduced immunogenicity, the rodent variable region still poses the risk of inducing an immune response. When it was recognized that the variable domains consist of a beta sheet framework surmounted by antigen-binding loops (complementarity determining regions or CDR's), humanized antibodies were designed to contain the rodent CDR's grafted onto a human framework. Several different antigen-binding sites were successfully transferred to a single human framework, often using an antibody where the entire human framework regions have the closest homology to the rodent sequence (e.g., Jones P T, et al., Nature 321, 522-5, 1986; Riechmann L. et al., Nature 332, 323-327, 1988; and Sato K. et al., Mol. Immunol. 31, 371-8, 1994). Alternatively, consensus human frameworks were built based on several human heavy chains (e.g., Carter P. et al., Proc. Nat. Acad. Sci. USA 89, 487-99, 1992). However, simple CDR grafting often resulted in loss of antigen affinity. Other possible interactions between the ∃-sheet framework and the loops had to be considered to recreate the antigen binding site (Chothia C, et al., Mol. Biol. 196, 901-917, 1987).
Comparison of the essential framework residues required in humanization of several antibodies, as well as computer modeling based on antibody crystal structures revealed a set of framework residues termed as “Vernier zone residues” (Foote J., et al., Mol Biol 224, 487-99, 1992) that most likely contributes to the integrity of the binding site. In addition, several residues in the VH-VL interface zone might be important in maintaining affinity for the antigen (Santos A D, et al., Prog. Nucleic Acid Res Mol Biol 60, 169-94 1998). Initially, framework residues were stepwise mutated back to the rodent sequence (Kettleborough C A, et al. Protein Engin. 4, 773-783, 1991). However, this mutation approach is very time-consuming and cannot cover every important residue.
For any particular antibody a small set of changes may suffice to optimize binding, yet it is difficult to select from the set of Vernier and VH/VL residues. Combinatorial library approaches combined with selection technologies (such as phage display) revolutionized humanization technologies by creating a library of humanized molecules that represents alternatives between rodent and human sequence in all important framework residues and allows for simultaneous determination of binding activity of all humanized forms (e.g. Rosok M J, J Biol Chem, 271, 22611-8, 1996 and Baca M, et al. J Biol Chem 272, 10678-84, 1997).
The above approaches utilize entire framework regions from a single antibody variable heavy or variable light chain to receive the CDRs. It is advantageous to provide highly homologous engineered antibodies based on antibodies from an originating species which exhibit reduced immunogenicity while maintaining an optimum binding profile that can be administered to a target species for therapeutic and diagnostic purposes.